Denoising Algorithms
Implemented algorithms
Below we list all methods currently implemented in our benchmark suite
| Name | Method name | Paper |
|---|---|---|
| CNN-10 | cnn10 | H. Chen, Y. Zhang, W. Zhang, P. Liao, K. Li, J. Zhou, and G. Wang, "Low-dose CT via convolutional neural network,” Biomedical Optics Express, vol. 8, no. 2, pp. 679–694, Jan. 2017 |
| RED-CNN | redcnn | H. Chen, Y. Zhang, M. K. Kalra, F. Lin, Y. Chen, P. Liao, J. Zhou, and G. Wang, “Low-dose CT with a residual encoder-decoder convolutional neural network,” IEEE Transactions on Medical Imaging, vol. 36, no. 12, pp. 2524–2535, Dec. 2017 |
| WGAN-VGG | wganvgg | Q. Yang, P. Yan, Y. Zhang, H. Yu, Y. Shi, X. Mou, M. K. Kalra, Y. Zhang, L. Sun, and G. Wang, “Low-dose CT image denoising using a generative adversarial network with wasserstein distance and perceptual loss,” IEEE Transactions on Medical Imaging, vol. 37, no. 6, pp. 1348– 1357, Jun. 2018. |
| ResNet | resnet | A. D. Missert, S. Leng, L. Yu, and C. H. McCollough, “Noise subtraction for low-dose CT images using a deep convolutional neural network,” in Proceedings of the Fifth International Conference on Image Formation in X-Ray Computed Tomography, Salt Lake City, UT, USA, May 2018, pp. 399–402. |
| QAE | qae | F. Fan, H. Shan, M. K. Kalra, R. Singh, G. Qian, M. Getzin, Y. Teng, J. Hahn, and G. Wang, “Quadratic autoencoder (Q-AE) for low-dose CT denoising,” IEEE Transactions on Medical Imaging, vol. 39, no. 6, pp. 2035–2050, Jun. 2020. |
| DU-GAN | dugan | Z. Huang, J. Zhang, Y. Zhang, and H. Shan, “DU-GAN: Generative adversarial networks with dual-domain U-Net-based discriminators for low-dose CT denoising,” IEEE Transactions on Instrumentation and Measurement, vol. 71, pp. 1–12, 2022. |
| TransCT | transct | Z. Zhang, L. Yu, X. Liang, W. Zhao, and L. Xing, “TransCT: Dual-path transformer for low dose computed tomography,” in MICCAI, 2021 |
| Trainable bilateral filter | bilateral | F. Wagner, M. Thies, M. Gu, Y. Huang, S. Pechmann, M. Patwari, S. Ploner, O. Aust, S. Uderhardt, G. Schett, S. Christiansen, and A. Maier, “Ultralow-parameter denoising: Trainable bilateral filter layers in computed tomography,” Medical Physics, vol. 49, no. 8, pp. 5107– 5120, 2022. |
Test set performance
Below we report the results of the best performing networks of each method on the test dataset. They can be reproduced by running python test.py --print_table (see Test models).
| Method | SSIM (Chest) | SSIM (Abdomen) | SSIM (Neuro) | PSNR (Chest) | PSNR (Abdomen) | PSNR (Neuro) | VIF (Chest) | VIF (Abdomen) | VIF (Neuro) |
|---|---|---|---|---|---|---|---|---|---|
| LD | 0.312 | 0.856 | 0.914 | 18.066 | 29.117 | 30.923 | 0.083 | 0.353 | 0.578 |
| cnn10 | 0.559 | 0.907 | 0.928 | 27.307 | 32.737 | 31.968 | 0.175 | 0.455 | 0.642 |
| redcnn | 0.584 | 0.913 | 0.932 | 28.002 | 33.685 | 34.132 | 0.205 | 0.504 | 0.715 |
| qae | 0.557 | 0.903 | 0.928 | 27.115 | 32.304 | 31.923 | 0.167 | 0.424 | 0.618 |
| wganvgg | 0.505 | 0.893 | 0.92 | 25.324 | 30.906 | 29.208 | 0.137 | 0.39 | 0.566 |
| resnet | 0.581 | 0.912 | 0.932 | 28.032 | 33.583 | 33.853 | 0.21 | 0.5 | 0.705 |
| qae | 0.557 | 0.903 | 0.928 | 27.115 | 32.304 | 31.923 | 0.167 | 0.424 | 0.618 |
| dugan | 0.544 | 0.904 | 0.93 | 26.316 | 32.468 | 32.078 | 0.156 | 0.441 | 0.656 |
| transct | 0.538 | 0.89 | 0.893 | 26.736 | 30.924 | 27.363 | 0.155 | 0.387 | 0.461 |
| bilateral | 0.529 | 0.871 | 0.905 | 25.057 | 27.357 | 29.238 | 0.143 | 0.373 | 0.541 |